Note: Descriptions are shown in the official language in which they were submitted.
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
SUN SENSORS USING MULTI-PINHOLE OVERLAYS FOR DETECTION OF SATELLITE ATTITUDE
FIELD OF THE INVENTION
The present invention relates to sun sensors and in particular to
sun sensors that provide the angular location of the sun with respect to the
sensor reference frame.
BACKGROUND OF THE INVENTION
Sun sensors are required equipment for most satellites since
they provide line-of-sight direction of the sun with respect to the satellite.
Such sensors are useful during several stages of the satellite mission
including transfer orbit; pre- and post- apogee motor firing; attitude
acquisition; momentum dumping; gyro calibration; yaw attitude sensing; loss
of attitude lock and re-acquisition; and other emergency and routine
requirements. Sun sensors of various types are also used for solar array
tracking, thermal control and sun presence detectors for star sensors and
earth sensors. Sun sensors are also useful for ground-based equipment that
require the location of the sun to be known.
The sun is a very bright source emitting throughout the visible
and infrared regions corresponding to a black body with a temperature of
about 5900K. This means that it has a peak in its spectral radiance at about
500 nm in the middle of the visible band. Thus most instruments that view
the sun use the visible region. Inevitably, much of the solar power falling
onto
such an instrument must be discarded to avoid overloading or overheating the
detectors and other portions of the instrument optics. However, the sun is so
intense that stray light issues can arise that remain significant even if
rejection
is high.
Given that the satellite is in orbit around the Earth, the location
of the sun changes continually with respect to the satellite body axes. For
equatorial orbits, the sun is constrained in elevation by the ecliptics.
However, for other orbits this is not the case and a broader field-of-view,
FOV, is required to continually monitor the sun's location. For the example
case of a geosynchronous telecoms satellite and considering an Earth-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
centred set of coordinate axes, the operational range of the sun will be ~23.5
degrees in elevation from the orbit plane. During the day, the sun will appear
to move through approximately 360 degrees in the azimuth plane. Normal
and emergency operations as indicated above will require occasional access
to a broader elevation range.
The accuracy required for a sun sensor is governed by its use in
the attitude control strategy of the satellite mission. For the
telecommunication satellite missions, the accuracy requirement is usually
derived from antenna pointing specifications which are in turn driven by gain
slopes attainable for the various radio frequency beams. As technology
advances and tighter beams are utilized for point-to-point communications,
the gain slopes are increasing requiring tighter tolerances on attitude
control.
Thus attitude resolution of an arc-minute or less is desirable and in certain
instances may be necessary.
The combined problems of wide angular FOV and high
resolution represent the major challenge to sensor designers. Most
instruments must compromise between these two contradictory requirements.
Having a FOV of 120x120 degrees with a resolution of 0.02 degree implies
6000 resolution elements in each direction. Thus multiple sensors are often
used to provide a wide field-of-regard, FOR, (multiple FOV's) while
maintaining the necessary resolution.
The sun subtends an angular size of 0.53 degree. For most
instruments the sun can be considered a point scurce. As higher resolutions
(such as 0.02 degree) are demanded this assumption must be re-examined
and its implications considered.
Manufacturers of digital sun sensors have relied on analog
technology for the sensing elements but superimpose elaborate masks to
provide the necessary resolution. Thus the sun at a given angle illuminates
certain sensing photocells through a main slit and a mask consisting of
reticle
slits. The digital signal is usually produced as Gray code information based
on which photocells produce voltages above a threshold level. This
technique is only effective in a single axis so that a two-axis sensor
consists
-2-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
of a pair of sensors mounted orthogonally.
This technology has limitations in regard to resolution owing to
the sensitivity of the photocells and the accuracy and alignment of the mask
slits. The sensors and the associated electronics are usually separate units.
A typical sensor head with an FOV of 128x128 degrees and a resolution of
0.25 degree has a volume of 130 cm3 and a mass of 260 g whereas the
associated electronics has a volume of 315 cm3 and a mass of 295 g. Thus a
system for full sky viewing consisting of five sensor heads and one
electronics
box has a total volume requirement of 965 cm3 and a total mass of 1595 g.
The power requirement for the system is about 120 mW. Higher accuracy
units require significantly more resources especially for the mass, volume,
and power for the processing electronics unit.
Thus it would be advantageous to prove a sun sensor that has a
smaller volume and a smaller mass.
SUMMARY OF ThlE INVENTION
The present invention provides an implementation of a sun
sensor that provides high resolution over a wide FOR. The sensor uses the
principles of a classic pinhole camera in conjunction with a modern two-
dimensional detector array. The sensor uses multiple pinholes located in a
dome-like housing over the detector array to provide multiple FOV's that
project onto a common array. The FOV's are positioned such that when
taken in combination comprise the required wide FOR. The overlay of
multiple sky images provided by the multiple pinholes onto the common
detector allows for the monitoring of the entire FOR since the sun is a unique
target within a background that is generally undifferentiated. Thus the sensor
takes advantage of this "overlay principle" to multiplex the many FOV's
required to retain a high resolution and a wide FOR simultaneously.
A position or sun sensor comprises a sensor housing, a plurality
of pinholes formed in the sensor housing, a detector mounted within the
housing and a method of processing the information detected. The detector is
mounted in the sensor housing. Each pinhole has a field of view and the
-3-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
detector receives the images from each field of view. Each field of view is
defined by the position of the pinhole relative to the detector. The images
are
received in an overlay relationship thereby providing a field of regard. The
processing method determines the presence and location of an object in a
field of regard.
Further features of the invention will be described or will
become apparent in the course of the following detailed description.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention will now be described by way of example only,
with reference to the accompanying drawings, in which:
Fig. 1 is a schematic drawing of the pinhole position geometry
used in the sun sensors of the p. esent invention;
Fig. 2 is a schematic drawing of the FOV's and FOR;
Fig. 3 is a schematic drawing of four adjacent FOV's;
Fig. 4 is a schematic drawing of the four FOV's of figure 3
imaged into a common detector array;
Fig. 5 is a perspective view of the sun sensor constructed in
accordance with the present invention;
Fig. 6 is a top view of the sun sensor of figure 5;
Fig. 7 is a side view of the sun sensor of figure 5;
Fig. 8 is a section view of the sun sensor taken along 8 - 8 of
figure 7;
Fig. 9 is a perspective view of the dome portion of sun sensor of
the present invention as viewed from the inside;
Fig. 10 a top view of an alternate embodiment of the dome of
the sun sensor showing uniquely shaped pinholes;
Fig. 11 is a schematic drawing of a uniquely shaped pinhole
relative to the detector axes;
Fig. 12 is a top view of another alternate embodiment of the
dome of the sun sensor showing pairs of pin holes;
Fig. 13 is a schematic drawing of a uniquely shaped pinhole
-4-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
relative to the detector axes; and
Fig. 14 is a perspective view of an alternate embodiment of the
sun sensor constructed in accordance with the present invention.
DETAILED DESCRIPTION OF THE INVENTION
Pinholes
The classic pinhole camera is the most fundamental of optical
instruments. The idealised pinhole camera consists of an extremely small
pinhole located above a film or detector plane. It entails no reflective or
refractive optics and the image is a true representation of the object,
although
inverted. Thus the geometric relationship between the detector array and the
pinholes) defines the extent and resolution of the FOV. Figure 1 illustrates
the pinhole principle shown generally at 10. The actual pinholes must be of a
finite size which is limited by considerations that will be discussed later.
The locations of the pinholes in the design are determined from
the FOV requirements and the detector size. The location of each pinhole 12
is defined by the intersection of the extreme rays 14 and 16 defining the
required FOV drawn from the corners of the detector array. Given the
symmetry of the FOV's, the pinholes are located in patterns around the
boresight axis (19). In operation the FOV is then physically defined by the
pinhole positions and the detector size.
Using a series of pinholes located using the method described
above, with respect to a common focal plane array 18, the required FOR is
divided into smaller FOV's corresponding to each pinhole. Thus the required
120x120-degree field becomes the FOR and the pinholes provide the
individual FOV's.
Figure 2 shows how a 124x124 degree FOR 30 can be divided
into twelve FOV's 20. The twelve FOV's are plotted as functions of azimuthal
angle and zenith angle on a polar plot. The pinholes for the four inner-FOV's
22 representing the central FOR are located such that central rays from the
detector array make an angle of about 24 degrees with the sensor boresight
(which is normal to the array). The outer group of four side-FOV's 26 that
-5-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
extend beyond the central region have their pinholes located on rays that are
about 51 degrees off the boresight. These are somewhat rectangular owing
to the projection of the square detector array via the side pinholes. The four
corner-FOVs 28 making up the corners of the full FOR have pinholes
positioned so that the central rays are about 60 degrees from the boresight.
As stated above the idealised pinhole camera has an infinitely
small pinhole. There are two reasons why the pinhole cannot be made too
small. Firstly the pinhole size helps determine the signal level generated by
the detector array pixels. Since the irradiance on the detector plane is
proportional to the pinhole area, the pinhole must be at least large enough to
meet the minimum signal to noise requirements. Secondly the size of the
pinhole is limited by diffraction effects. The smaller the pinhole is, the
larger
the diffraction effects are. Some diffraction is unavoidable and acceptable,
but large diffractive effects would be required to be well understood and
controlled. Diffractive effects in their simplest form would limit the
resolution of
the image.
However, increasing the pinhole size also limits the resolution of
the image since the pinhole has a geometric extent with an effect that can be
thought of as a blur spot. The effective energy distribution on the focal
plane
is therefore a convolution of the sun's image, the geometric projection of the
pinhole, and the intensity distribution resulting from diffraction.
To assess the size of these various contributions it is best to
consider a specific example. For a typical detector of 1 cm square, the sensor
dimensions are such that the sun's image is only about 100um in diameter.
For a 100pm diameter pinhole, the blurring corresponds to100um microns
and the central lobe of the Airy diffraction pattern is also on the order of
100Nm for visible light. Therefore each of these effects would be of equal
importance. Increasing the pinhole size would decrease the diffractive effects
but increase the size of the pinhole's geometric projection. The opposite is
true if the pinhole size is reduced. However, because it is only the centroid
of
the image that is important and not a resolved image of the sun, the relative
contribution each effect has is not critical. Therefore there is some latitude
in
-6-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
choosing the pinhole size and the size of the convolved image on the detector
can be optimised.
Overlay Principle
The function of the sun sensor is to determine the location of
the sun within a specified coordinate system with sufficient accuracy and at a
required update rate. Given that the sun is a unique object in the sky and
that
the background features can be readily ignored or easily discriminated, the
sun sensor requirements represent an ideal application of the "overlay
principle". This technique involves the superimposition of several fields-of-
view onto a single detector array as illustrated in Figures 3 and 4 where
Figure 3 depicts four separate fields-of-view that are combined in Figure 4 on
the common detector array. This principle may be used where unique targets
exist within a generally undifferentiated or readily characterisable
background.
In general, this approach can lead to instruments that offer high resolution
while providing a large field-of regard. This capability is usually considered
inconsistent but the "overlay principle" provides a limited solution to this
general problem.
The sensor concept provides for twelve FOV's 20 to cover a
124x124 degree FOR 30 with the inner 22 and corner 28 FOV's
corresponding to about 33x33 degrees in locally centred coordinates and the
side 26 FOV's corresponding to about 67x33 degrees in locally centred
coordinates. Each FOV includes a 2-degree overlap region around its
periphery so that transitions from one FOV to the next can be continuously
tracked.
There is an inherent degeneracy in the position coordinates of
the sun in this technique, that is, the absolute location of the sun is not
known
without knowing which field-of-view provided the image with the sun in it.
There are several means for resolving this degeneracy relying on either
pinhole shape, multiple pinholes, or a history of satellite position.
The locations of the pinholes in the design are determined from
the FOV requirements and the detector size. The location of each pinhole 12
-7-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
is defined by the intersection of the extreme rays 14 and 16 defining the
required (FOV) drawn from the corners of the detector array. Given the
symmetry of the FOV's, the pinholes are located in patterns around the
boresight axis (19). In operation the FOV is then physically defined by the
pinhole positions and the detector size. The following table lists the
positions
calculated for a 1 cm by 1 cm detector and a 120 degree by 120 degree FOR.
The radial distance and the angles phi and theta are respectively the
standard radial, elevation and azimuthal spherical coordinates. The
coordinates are referred to the centre of the detector with theta=0
corresponding to one of the detector axes.
Pinhole NumberRadial distancePhi [degrees] Theta [degrees]
[mm]
1 12 39 0
2 12 39 90
3 12 39 180
4 12 39 270
5 15 30 45
6 15 30 135
7 15 30 225
8 15 30 315
9 16 66 45
10 16 66 135
11 16 66 225
12 16 66 315
Detector and Electronics
The detector array and its integrated processing electronics of
the present sun sensor represent a departure from the prior art. Preferably
the detector is fabricated in a CMOS process as an active pixel sensor.
Preferably the sensing element is silicon and the circuitry for reading out
each
pixel is fabricated in the same process within each pixel cell. These
detectors
_g_
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
operate in the visible spectrum although there is no reason why detectors
sensitive to other wavelengths cannot be used with this sensor design.
Typically an active pixel sensor includes double correlated sampling, analog-
to-digital conversion, row/column selections, and random access. The
detector operation is much different from the traditional charge-coupled-
device (CCD) as charge is not shifted over the array but read out directly.
Since the detector is fabricated in CMOS, other processing
elements can be added on the same chip thus providing the required
integration of detector and processing. However, in order to minimize
development risk, cost, and schedule, the microprocessor and associated
memory elements may be incorporated with the detector as part of a multi-
chip module (MCM). This arrangement may provide all of the required
processing although the architecture and the distribution of processing
between the detector chip and the microcontroller is a matter for
optimization.
The processing may include detecting the sun presence in a particular region
of the array, reading out the pixels in the vicinity, performing the necessary
centroiding and weighting operations to determine the sun position, and
converting the pixel coordinates to a sensor-based coordinate system.
Finally, the MCM microcontroller may interface with the satellite computer to
pass on the sun position information at the required update rate. Many
modern satellites have excess processing capacity within their on-board
control computers. Thus, it may be effective to allocate the processing of the
sun sensor to the central computer and have the sensor produce raw or semi-
processed data periodically or on request.
Although a typical sun location update rate required by a
satellite is 10 Hz, with on chip processing the detector may be read out at
higher rates such as 100 Hz. This would enable the possibility of two modes
of operation. A normal mode would use central computing for optimal
accuracy while a second emergency mode with on chip processing could
achieve higher readout rates at the expense of accuracy. The emergency
mode could be employed, if for example the satellite were spinning out of
control.
_g_
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
Packa ing
Referring to figures 5 - 9 the sun sensor is shown generally at
39. The sensor concept lends itself to a very compact design with most of the
volume contained in the pinhole dome. Preferably sun sensor 39 includes a
dome 40 consisting of a hollow hemispherical piece of glass with an inner
surface 42 a predetermined distance to the centre of the detector surface.
This distance defines the distance from the closest pinholes to the detector
surface. For a detector which is 1 cm on each side, the radius of the sphere's
inner surface is slightly less than 12mm. The radius of the sphere's outer
surface is slightly larger than 16mm. Therefore the minimum thickness of the
dome wall has been calculated for this example to be 5mm. The pinholes are
located in bore holes 44 that extend outwardly from the inner surface 42 of
the dome 40 but do not extend through to the outer surface 46 thereof. It will
be appreciated by those skilled in the art that the position of the pinhole in
the
thickness of the dome will vary depending on the desired field of view as
described above. Accordingly, the depth of each bore hole 44 will be
adjusted to position the pinhole at the desired depth. Preferably each pinhole
is photo-etched or lithographically deposited onto the glass so that no
physical pinholes would be required. Preferably the outside surface 46 of the
dome 40 is coated so as to attenuate the solar input, provide thermal
reflection, and protect against electrostatic buildup. A thin gold coating is
ideal
for an application in space. The gold will reflect most of the thermal
radiation
and the thickness of the gold can be selected to be such that the correct
irradiance passes through the pinhole and onto the detector. Preferably the
glass dome is mounted onto an aluminum base 48. Similarly the inner
surface between the boreholes is coated so that light does not pass through
the dome between the boreholes.
Alternatively, the dome may be made from a metal or plastic
substrate with pinholes inserted into the appropriately machine recesses. For
solar applications typically aluminium or titanium would be used because of
their weight and thermal characteristics. Whereas for ground based systems
plastic or other metals may be used. Figure 14 shows generally at 94 a
-10-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
housing made from an opaque solid. The depths of the recesses 96 are such
that the pinholes 98 are located the calculated distance from the detector as
described earlier. The angle of the conical recess is selected to match the
FOV for that particular pinhole. Therefore the shape and the slope of the
recess is dependent on the position of the pinhole in the dome. Boreholes as
described above or formed in the inner surface of the dome in registration
with the pinholes as described above. The pinholes themselves may be
conventional, mechanical pinholes manufactured from a thin metal foil or they
may be photoetched or lithographically deposited onto a transparent
substrate that are positioned in the pinholes to fill them. The latter option
would allow for the pinhole to simultaneously act as a filter to attenuate the
solar throughput.
The pinholes may be made in various non-circular shapes to
eliminate degeneracies; for example an L- or T- shaped pinhole may be used
in various orientations. Relying on shaped pinholes requires that they be
rather large so as to dominate the image shape when convolved with the
other effects. In the example above the size of the pinhole would have to be
larger than 100Nm. Similarly, the shape of the pinhole may be used for
substantially smaller pinholes but then it may be the diffraction pattern not
the
actual pinhole shape that would distinguish the pinholes.
Figures 10 and 11 illustrate an example of a sun sensor shown
generally at 60 using pinholes having predetermined shapes and unique
orientations. The shapes shown herein are T's and L's. The shapes and
orientations of the pinholes are such that each pinhole has a non-degenerate
image. The sizes of the pinholes shown herein are greatly exaggerated for
clarity of presentation. Firstly the shape of the resultant spot on the
detector is
determined by a simple algorithm. An example of a simple algorithm is one
that determines the centroid of the image to differentiate the T's from the
L's.
Symmetry requires that the centroid of a tee must lie within the image of the
tee itself. Whereas, in the case of an L the aspect ratio of arm length to
slit
width can easily be chosen such that the centroid does not lie within the
image. Hence the T's and L's can be distinguished by a simple algorithm that
-11-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
checks whether or not the centroid lies within the image. The angle 62 the
shape makes with the detector axes 64 then identifies which T or L is forming
the image. In the example shown herein the centre pinholes 66 are T
shapes, the side pinholes 68 are T shapes and the corner pinholes 70 are L
shapes.
It is also possible to use two pinholes located in close proximity
to each other so as to operate as a paired set of pinholes. If the pinholes
are
mounted such that the line connecting the pinholes is perpendicular to a line
normal to the detector surface, then the distance separating the two pinhole
images is preserved regardless of sun angle. Having pairs of pinholes with
varied separations can therefore resolve any degeneracies. Note that this
technique requires the overlap of the FOV's to be large enough to completely
contain the largest pinhole pair.
Figures 12 and 13 illustrate another example of a sun sensor
shown generally at 72 showing the use of two pinholes to non-ambiguously
identify which pin holes formed the image. As above, the size and spacing of
the pin hole pairs has been greatly exaggerated to improve the clarity of
presentation. The centroids of the two pin holes define a line 74. The angle
76 the line makes with the detector axes 64 identifies to which of four
symmetry axes the pin holes belong. The length of the line 78 corresponds to
the spacing of the pin holes and uniquely identifies which pin hole on the
symmetry axis is casting the image. The central pinholes 80 and the corner
pinholes 82 are arranged along a first 84 and second 86 orthogonal axes.
Each pinhole along each of the first and second axes has a unique
predetermined spacing. Similarly the side pinholes 88 are arranged along a
side first 90 and second 92 orthogonal axes and each pinhole along each
side first and second axes has a unique predetermined spacing.
Finally, the history of the satellite position can also determine
which pinhole is illuminated. Since there exist non-degenerate cases where
the images lie in the regions of FOV overlap, these positions can be used to
initiate a tracking of satellite position. From these non-degenerate positions
the evolution of the satellite position can be determined so long as the
-12-
CA 02391873 2002-05-15
WO 01/38896 PCT/CA00/01350
changes in satellite orientation from one measurement to the next is
sufficiently small. Thus this technique may not be useful for a satellite
spinning rapidly as it might during emergency manoeuvers or when out of
control.
Thermal sensitivity could potentially limit the ultimate accuracy
of the sensor. Efforts taken to minimise the sensor's temperature variability,
such as a low emissivity coating on the outside of the dome and pinholes,
could only reduce but not eliminate thermal changes. These thermal effects
could be reduced by selecting materials for manufacture such that the relative
positions of the pinholes with respect to the detector would change only
minimally. Further improvements could be had by actively stabilising the
sensor temperature or by calibrating the effects of thermal changes.
Referring to figures 5 and 8 the MCM is mounted to the base of
the sensor under the dome on a printed circuit board 50 and any additional
electronics for power conditioning and/or satellite telemetry interface are
mounted in the region around the chip or on the other side of the PCB. It is
expected that the sensor would be mounted to a satellite with three bolts via
a
flange or tabs 52 arranged around the periphery of the unit. The satellite
interface connector providing power and data interfaces may be mounted
beneath the sensor with a cut-out in the satellite panel to accommodate the
interconnection cable. A bonding lug may be provided in the base of the
sensor so that the unit may be strapped as required to the satellite
structure.
Several areas around the periphery of the sensor base may be machined flat
and polished so that they may be used as sensor alignment reference
surfaces.
It will be appreciated that the above description relates to the
invention by way of example only. Many variations on the invention will be
obvious to those skilled in the art and such obvious variations are within the
scope of the invention as described herein whether or not expressly
described.
-13-
